Nobody has a hard and fast answer to this and the recommendations differ somewhat. Some experts advise an early start (as soon as the viral loading exceeds 5,000-10,000 copies/ml) regardless of the CD4-cell count. This means that the great majority of patients should be treated. Is this desirable? Arguments based on virology and immunology (with the necessary uncertainty and regular new information) sometimes conflict with the clinicians' findings (also with their uncertainties, subjectivity and doubts). HIV is a chronic infection and the question arises whether patients can take these cocktails for 10 or 15 years and tolerate the side-effects for all this time. Compliance (or non-compliance) with therapy is a very important issue. Furthermore, there is the cost of all this. On the other hand experience with the drugs is limited in time and new clinical and therapeutic data are constantly appearing. Tritherapy is usually started in symptomatic patients, patients with less than 350 CD4 cells/l and patients with a viral load of more than 30,000 copies/ml. However, these criteria serve only as guidelines. They should be individually adjusted and a decision made for each individual patient. Antiviral treatment is potentially dangerous for the foetus in the first three months of pregnancy.

In developing countries it is best to wait before starting antiviral therapy, certainly if the patient has insufficient financial means to continue the treatment. Moreover, protease inhibitors are difficult to combine with the classical antituberculosis treatments. The instability of certain products in high temperature environments should be taken into account (e.g. Kaletra).